Article Text

Download PDFPDF

Predicted mortality from malignant mesothelioma among women exposed to blue asbestos at Wittenoom, Western Australia
  1. A Reid1,
  2. G Berry2,
  3. J Heyworth1,3,
  4. N H de Klerk1,4,
  5. A W Musk1,5
  1. 1
    School of Population Health, University of Western Australia, Crawley, Western Australia, Australia
  2. 2
    School of Public Health, University of Sydney, Sydney, New South Wales, Australia
  3. 3
    Faculty of Medicine, Dentistry and Health Sciences, University of Western Australia, Crawley, Western Australia, Australia
  4. 4
    Telethon Institute for Child Health Research, and Centre for Child Health Research, University of Western Australia, Subiaco, Western Australia, Australia
  5. 5
    Department of Respiratory Medicine, Sir Charles Gairdner Hospital, Nedlands, Western Australia, Australia
  1. A Reid, Occupational Respiratory Epidemiology, School of Population Health, M431, University of Western Australia, 35 Stirling Highway, Crawley, WA 6009, Australia; alison.reid{at}


Introduction: Nearly 3000 women and girls were documented to have lived at the blue asbestos mining and milling town of Wittenoom in Western Australia between 1943 and 1992. Eight per cent of deaths among these women to the end of 2004 have been from malignant mesothelioma of the pleura.

Aim: To predict future mortality from mesothelioma to 2030 in this cohort.

Methods: Mesothelioma mortality rates incorporating parameters for cumulative exposure, a power of time since first exposure and annual rates of fibre clearance from the lung were calculated from maximum likelihood estimates. These rates plus age specific mortality rates for Western Australian females incorporating an excess lung cancer risk were then applied to all Wittenoom cohort women surviving to the end of 2004, in yearly increments, to predict the future numbers of cases of mesothelioma to 2030.

Results: There were 40 deaths from mesothelioma among the Wittenoom women to the end of 2004. Using a range of models that incorporate time since first exposure, competing risks from other diseases, latency periods and clearance of mesothelioma from the lungs we predict 66 (lowest estimate) to 87 (highest estimate) deaths from mesothelioma until 2030. This represents one and a half to two and a half times the number of deaths that have already occurred to the end of 2004.

Conclusion: The high toll from mesothelioma in this cohort of women and girls will continue well into the future.

Statistics from

Mesothelioma incidence traces historical asbestos consumption.1 2 The future burden of mesothelioma has been predicted in cohorts of asbestos exposed workers35 and among whole populations of workers in occupations with asbestos exposure.6 Among the Wittenoom miners and millers (mostly men), de Klerk et al predicted 692 cases of mesothelioma between 1987 and 2020.3 Berry, using the same cohort, predicted 255–654 mesotheliomas for the same period4 and updated this prediction in 2004, estimating 110 deaths between 2001 and the end of 2020.7

The predictions from Wittenoom have predominantly focussed on men who were exposed to higher levels of asbestos through their employment. However, 3000 women and girls were documented to have lived in and worked at the blue asbestos milling and mining town of Wittenoom between 1943 and 1992. The aim of this study is to predict future mortality from malignant mesothelioma among women and girls who were exposed to blue asbestos at Wittenoom, using quantitative measures of intensity and duration of exposure that have been previously estimated on these women.8 9


Women and girls at Wittenoom

Approximately 3000 women and girls (41% were aged less than 15 years on arrival) were documented to have lived in and worked at the blue asbestos milling and mining town of Wittenoom, in the remote Pilbara region of Western Australia. Most came to Wittenoom as the wives or daughters of an Australian Blue Asbestos Company (ABA) miner or miller, or worked as teachers, nurses or service providers for the state government.10 11

Women ABA workers

From employment records, 416 women were identified as working for the ABA over the period 1943–1966. Mostly they were not employed in mining or milling roles but instead worked in the company shop, hotel or offices. The vital status of 58% of the women was known when the cohort was assembled in the 1970s from the ABA employment records.8 Intensive and continuous follow-up since that time using various methods including state and commonwealth electoral rolls, birth, marriage and death registers, state and national cancer registries and electronic white pages, had established the vital status of more than 70% of the former female workers by the end of 2000.

Female residents of Wittenoom

Former male and female residents (eg, those not employed directly in asbestos mining or milling – non-ABA workers) of Wittenoom12 13 were identified from several sources: state primary school records (22%), admission and out-patient records of the Wittenoom hospital and general practitioners (20%), participants in a cancer prevention program14 15 and associated publicity (18%), questionnaires sent to ABA workers (14%), the state electoral roll for the Pilbara district (12%) and Wittenoom birth records (4%). Other sources included records from the Catholic Church, Wittenoom burial records, employment lists from the school, hotel, police, hospital and banks, and information from the Asbestos Diseases Society of Western Australia (10%). A total of 5097 individuals were identified from 18 553 records.9 12 Between 1991 and 1993 a questionnaire was sent to all former residents of Wittenoom traced to an address in Australia (n = 3244; 64%), except for those participating in a cancer prevention program (n = 641; 13%) from whom the information had already been collected. Date, length and place of residence at Wittenoom, occupation at Wittenoom, whether lived with an asbestos worker or washed the clothes of an asbestos worker, smoking and past medical history as well as demographic information were collected.16

Follow-up status at the establishment of the cohort in 1993 was: 47% had returned a questionnaire, 14% were participating in the cancer prevention program, 17% had not returned a questionnaire, 10% were dead, 11% had not been traced since leaving Wittenoom and 1% had permanently departed from Australia.13

To make the residents’ data as complete as possible, some assumptions were made where the person remained untraced, did not return a questionnaire or was dead. If they were related to an ABA worker, then dates and place of residence were assumed to be identical to those of the worker. For those unrelated to an ABA worker, dates of residence were assumed to be the same as for other family members provided at least one family member had known exposure. Dates of residence were taken as those found on the various sources used to establish the cohort for all other residents.9 13 If the untraced person was the wife of an ABA worker and was known to have lived with that worker, it was assumed that she had washed his clothes. The residents’ cohort was considered complete when comparisons between it and the population of Wittenoom recorded at various census dates showed a close correspondence.12

Work has continued on the development of this cohort since 1993 and this accounts for differences in the number of persons from earlier papers.9 12 13 To the end of 2000 there were 2608 women in the residents’ cohort.17 The mortality of these environmentally exposed women has been previously assessed.10

Asbestos exposure measurements

Asbestos exposure measurements have been defined elsewhere for former workers8 and residents.9


Numerous measurements of dust concentration using a koniometer were taken in the mine and mill between 1948 and 1958 (when a new cleaner mill was established), and in 1966 a survey of fibre counts using a Casella long running thermal precipitator was undertaken across the industry and at various sites around the town.8 Estimated fibre concentrations ranged from 20 f/ml in the mine to 100 f/ml in the bagging room. The ABA employment records, contributions to a mine workers relief fund for workers prior to 1943 and the measurements taken during the surveys enabled cumulative exposure in f/ml-years to be calculated for each worker, by adding over all of their jobs the product of the length of time in the job and the estimated fibre concentration related to that job. An additional amount was added to their occupational exposure to account for 16 h of daily residential exposure and a 2-day weekend. This was calculated according to the residential exposure methodology (described below) for the period they were at Wittenoom. For example, an ABA worker who worked at Wittenoom for 60 days in 1965 had a cumulative exposure of 14.78 f/ml-years from their occupational exposure. An extra 0.055 f/ml-years (0.5/365.25)×((60×16)/24) was added to their cumulative exposure to account for their residential exposure.


Environmental monitoring throughout the town and surrounds occurred in 1966, 1973, 1977, 1978, 1980, 1984, 1986 and 1992.9 Based on the surveys that used personal monitors, residents of the township (not working directly with asbestos) were assigned an intensity of exposure of 1.0 f/ml from 1943 to 1957 and 0.5 f/ml between 1958 and 1966 when the mine and mill were closed. Interpolation between surveys established exposures of 0.5 f/ml in 1966 to 0.010 f/ml in 1992. Duration of residence was extracted from a questionnaire sent to each former resident or was taken as that found on documentation used to establish the residents’ cohort. Cumulative exposure was derived by summing over the years of residence, the product of fibre concentration for each year of residence and the length of time spent in Wittenoom during that year.9

The estimates of asbestos exposure have been shown to be internally valid, demonstrating an agreement with lung fibre burdens18 and an exposure–response relationship for all asbestos related diseases.8 13 1921 Further, Hodgson and Darnton found the Wittenoom exposures comparable to those exposures reported from other crocidolite mines and the risk of lung cancer among the Wittenoom workers was similar to that in other studies.22

Case ascertainment

Fifty six women were excluded because they had insufficient identifying information (missing date of birth, first name, etc) or because they were residents of Wittenoom for less than 1 month. Therefore, 2968 women (416 workers and 2552 residents) were followed up at the Western Australian Cancer Registry, the Western Australian Mesothelioma Register and the Australian Mesothelioma Register for notification of mesotheliomas. The Western Australian Registrar General was contacted for deaths occurring in Western Australia to the end of 2004. The National Death Index and the National Cancer Statistics Clearing House through the Australian Institute for Health and Welfare were searched for deaths and cancer incidence occurring in all other Australian states to the end of 2000. Women are difficult to follow up over long periods of time, given their changes of surname upon marriage, divorce and remarriage. Despite the use of these national resources, 27% of the women were lost to follow-up at the end of 2004.

Statistical analysis

Person-years at risk were calculated assuming that all women not known to be dead and not known to have migrated were alive at the end of 2004 (or if they were residents of other Australian states alive until September 2000). This assumes that all those who were lost to follow-up remained alive to the end of follow-up and is therefore likely to overestimate actual person-years at risk (and minimise the rate estimate).

Mesothelioma rates, per 100 000, were derived by dividing the number of deaths for each category of cumulative exposure or time since first exposure by the person-years at risk and multiplying by 100 000.

Derivation of mesothelioma mortality rate models

It has been established that mesothelioma mortality rates are related to time since first exposure raised to a power k, a constant, lying between 3 and 4.4 23 24 This relationship is modelled as rate = ctk where c represents cumulative exposure. When time since first exposure is lagged (to account for the time between initial malignancy and diagnosis or death), the relationship is modelled as rate = c(t−w)k where w is the lag period. These models hold where every increase in exposure represents continuous exposure to the carcinogen. This was considered reasonable for amphibole asbestos fibres because they are retained in the lung for very long periods after exposure. However, there is also evidence that asbestos fibres, including the amphiboles, are cleared from the lungs. This has been observed in experiments in rats25 and baboons26 as well as from lung fibre counts on lung tissue at post mortem among occupationally exposed cohorts.18 Lung fibre counts at post mortem from female gas mask workers exposed to a high intensity but relatively short duration of crocidolite asbestos suggested a rate of clearance of approximately 15% per annum.24 Further, epidemiological evidence suggests that mesothelioma mortality rates may begin to decline in occupationally exposed cohorts after 50 years since first exposure.27 28

To include a parameter for the rate of asbestos clearance, the above models have been amended to: mesothelioma rate = (ce−λt(t−w)k where λ is the rate of clearance of crocidolite from the lungs.24 Maximum likelihood estimates for the parameters of c, k and λ for each of the models were calculated using Stata v 9 (Stata, College Station, TX). Correlation between the λ and k made joint estimation of these parameters difficult, therefore clearance rates of 6.7% and 15% per year were fixed and a generalised linear model with a Poisson distribution was used to estimate k only. Elimination of 6.7% per year equates to a half life of 10 years.24 A model incorporating 15% elimination per year and a 5-year lag closely approximated the observed rate of mortality among the male Wittenoom workers.7

Predicting future mesothelioma mortality

To predict future mortality, the method devised by Berry4 in 1991 was used and various assumptions were made:

  • All cause mortality for the Wittenoom women was assumed to be the same as for Western Australian women for the period 2000–2004. Work in progress showed the standardised mortality ratio (SMR) for all cause mortality was 0.98 (0.90–1.06) assuming those women who were lost to follow-up were alive at the end of 2004.

  • Mortality from lung cancer was higher among the Wittenoom women than women in the Western Australian population. Therefore, the Western Australian lung cancer mortality rate for the period 2000–2004 was multiplied by the SMR for lung cancer (for the Wittenoom women) for categories of cumulative exposure.

The probability of dying in a year, averaged over each age, was applied to the women not known to be dead at the end of 2004. This gave the number of women who survived to the end of 2005. The probability of dying in a year was then applied to the surviving women of 2005 which gave the number of surviving women up to the end of 2006. And so on, until 2029. The deaths for each year were calculated by subtracting the number of survivors in year x+1 from year x. The annual number of mesothelioma deaths was calculated by multiplying the mesothelioma death rate, averaged for each age, by the number of surviving women in each year.


Mesothelioma mortality

There were 40 deaths from malignant mesothelioma among the Wittenoom women between 1950 and 2004 (10 among workers and 30 among residents). The mortality rate for workers was greater than that for residents for every period of time since first exposure. The rate appears to peak for workers at 30 or more years since first exposure, but for residents the rate is still increasing, with the rate at 40 or more years since first exposure 43% greater than the rate for 30–39 years since first exposure (table 1). Mesothelioma rates tended to increase for every increased category of cumulative exposure for residents, but the pattern was not as clear for workers. Cumulative exposure was derived differently for workers and residents (see Methods).

Table 1 Mesothelioma mortality rates, per 100 000 person-years*, by cumulative exposure and time since first exposure among Wittenoom women, 1950–2004

The number of mesotheliomas and person-years at risk by categories of time since first exposure and cumulative exposure are shown separately for residents and workers in table 1 of the online supplement. Few women have contributed person-years to the later periods of time since first exposure. The median time since first exposure for residents was 40.6 years (inter-quartile range (IQR) 33.7–46.1) and for workers 40.8 years (IQR 35.9–44.6). The values used to calculate maximum likelihood estimates of c and k, for the mesothelioma rate models, using Poisson regression are shown in table 1 of the online supplement.

Maximum likelihood estimates of c and k and parameters for categories of cumulative exposure and worker or resident status for the nine mesothelioma rate models are shown in table 2. The model with zero lag and zero clearance estimated a power (k) of 3.00. Inclusion of a lag period reduced the power to 2.51 with a 5-year lag and 2.01 with a 10-year lag. Using the parameters shown in table 2, the mesothelioma rates were then estimated for each woman using the equation: mesothelioma rate = exp[a+b+k*ln(t−lag)] without clearance and mesothelioma rate = exp[a+b+k*(t−lag)−0.067t] or mesothelioma rate = exp[a+b+k*(t−lag)−0.15t] for 6.7% and 15% clearance rates, respectively.

Table 2 Parameters (SE) for mesothelioma rates, using Poisson regression, and equations used to estimate future mortality from all causes and malignant mesothelioma among the Wittenoom women

The observed pattern of mesothelioma for the period 1950–2004, fitted to the mesothelioma rate models with zero, 5- and 10-year lag periods and a zero lag period and 6.7% and 15% clearance rates per annum is shown in fig 1.

Figure 1 Fitted mesothelioma rates over average exposure.

The models fit the data equally well up until about 45 years since first exposure, which covers most of the observed follow-up. At 40 years since first exposure the mesothelioma rate is approximately 150 per 100 000 person-years which concurs with the observed rate in table 1. The models diverge quite markedly after 45 years since first exposure given the fewer observations. Introducing a clearance rate into the mesothelioma model reduces the mortality rate substantially. The model with zero lag reaches 550 per 100 000 person-years at 60 years since first exposure. When 6.7% and 15% clearance rates are incorporated into the model, the rates reduced to approximately 360 and 220 per 100 000 person-years. Introducing a lag period into the model also reduces the mortality rate. Compared with the zero lag at 60 years since first exposure, the mortality rate with a 5-year lag reduces to 500 and further reduces to 450 with a 10-year lag. Figure 1 demonstrates the wide range of estimates obtained from the various mesothelioma models given the various clearance and lag scenarios. All of these models will be used to predict mesotheliomas to cover the wide range of possible future scenarios.

Parameters and equations used to predict future mortality are listed in table 2.

Overall, 2402 Wittenoom women were not known to be dead as at the end of 2004. This included 650 (27%) women who were lost to follow-up. Twenty per cent of the women were aged less than 45 years in 2004. The median age for workers was 68 years (IQR 62–78) and residents 55 years (IQR 46–67) (see table 2 of the online supplement).

Mortality from all causes and malignant mesothelioma for the period 2005–2030 for the nine mesothelioma rate models are shown in table 3. Around 49% of Wittenoom women are predicted to die from all causes over the 25-year period. Model 1 (zero lag) estimated the highest number of deaths from all causes (1181) and the greatest number of deaths from mesothelioma (87). Including a 5-year lag period slightly reduced the number of deaths (1179 deaths, 85 from mesothelioma), with a marginally greater reduction for the model with a 10-year lag (1178 deaths, 83 from mesothelioma). Including clearance rates in the models with zero lag predicted fewer mesotheliomas at 79 and 72 for 6.7% and 15% clearance rates, respectively. Including clearance rates in models with 5- and 10-year lags further reduced the number of deaths from mesothelioma and all causes. Model 9 which included a 10-year lag and 15% clearance rate predicted the fewest deaths from all causes (1167) and mesothelioma (66). This lowest number of predicted mesotheliomas is over one and a half times greater than the number of mesothelioma deaths that have already occurred to the end of 2004 (n = 40).

Table 3 Predicted mortality from all causes and malignant mesothelioma for all mesothelioma rate models for the period 2005–2030

The distribution of deaths from all causes and mesothelioma, in 5-year periods and 5-year age groups, as predicted by model 6 incorporating a 5-year lag and 15% clearance rate are shown in table 4. This was the model that most closely predicted mortality to that observed among the male Wittenoom workers.7 Deaths from all causes among the Wittenoom women steadily increase over this time span from 190 deaths in 2005–2009 to 308 deaths in 2025–2030. The number of mesothelioma deaths are relatively stable over this time period, increasing slightly from 12 deaths in 2005–2009 to 16 deaths in 2025–2030. The age distribution of mesotheliomas is similar to the pattern that was observed up to the end of 2004, in which the median age at death was 60 years (IQR 52–75).

Table 4 Age and period distribution of predicted mortality from all causes and mesothelioma for mesothelioma model 6 with a lag of 5 years and elimination of 15%


There have been 40 deaths from malignant mesothelioma among the women of Wittenoom between 1950 and 2004. Using models that incorporate competing risks from other causes, time since first exposure, lag periods and rates of clearance of crocidolite from the lungs, we predict that to 2030 there will be between 66 and 87 deaths from malignant mesothelioma among 2402 women not known to be dead at the end of 2004. This is over one and a half and up to two and a half times the number of mesothelioma deaths that have occurred to date. The number of deaths each year is not predicted to decline over the period to 2030.

Including a parameter for fibre clearance in the mesothelioma model attenuated the predicted number of mesothelioma deaths. In an earlier comparison between observed and predicted numbers of mesotheliomas among East London factory workers, not including a rate of clearance tended to overestimate the predicted number of mesotheliomas. For the period 1976–80, 17 mesotheliomas occurred compared with a predicted range of 28–37, although in the period 1972–1975, 23 mesotheliomas occurred compared with a predicted range of 17–22.29 Evidence suggests that between 10% and 15% lung fibre clearance occurs per annum in humans.24 Among the male Wittenoom workers, followed up to 2000, the pattern of observed mesotheliomas was most closely estimated by a mesothelioma model that included a 5-year lag and a 15% annual clearance rate.7 If the pattern observed among the male workers holds true for the Wittenoom women, then a further 69 deaths from mesothelioma may occur up to 2030. Berry estimated that clearance of 15% per year occurred among female gas mask workers with heavy but short exposure.24 The Wittenoom women obtained their asbestos exposure largely from their environment (tailings from the mining process were distributed throughout the town) and those with occupational exposure were generally not exposed in the mine or mill.11 To our knowledge this is the only study that has predicted future mortality from mesothelioma among women who have been environmentally exposed to asbestos. Inhaled fibres resulting from environmental exposure are shorter and finer and therefore more rapidly removed from the lungs than longer fibres.30 As a result, lung clearance among the Wittenoom women may have occurred at rates faster than 15% per annum. However, Baraldo et al suggested that women have a smaller lung volume but a higher forced expiratory flow rate than men and therefore a higher ratio of FEV1 to FVC,31 which may lead to greater fibre retention than in men with larger lungs.32 If this was the case we might expect a higher risk of mesothelioma in women at the same levels of exposure as men, yet this has not been observed.17

Until the end of 2004 the mesothelioma mortality rate appeared to be increasing for the Wittenoom women, although there was some evidence that the maximum rate may have been reached for the women workers. This pattern also holds for the predicted mesotheliomas. Absolute numbers of predicted mesotheliomas are relatively stable over the whole period 2005–2030, with no obvious decline. Among male Wittenoom workers, Berry predicted a peak of nine cases per year occurring between 2001 and 2005, six cases per year between 2006 and 2010, and four cases per year between 2011 and 2015 falling to two cases per year between 2016 and 2020, for the model with a 5-year lag and 15% elimination.4 We earlier reported a longer latency period for Wittenoom residents compared to ABA workers.17 A longer latency period appears to be consistent with lower exposures to asbestos and therefore a lower risk for mesothelioma. A longer latency period among the women may explain why there has been no obvious decrease in absolute mesothelioma predicted numbers to 2030.

Main messages

  • We predict 66–87 mesotheliomas between 2005 and 2030 among 2402 women exposed to crocidolite environmentally and occupationally at Wittenoom.

  • This is over one and a half and up to two and a half times the number of mesotheliomas that have already occurred to the end of 2004.

Far fewer cases and lower rates of mesothelioma have occurred among the Wittenoom women compared with the (largely male) asbestos miners and millers cohort, with 202 cases of pleural and 33 cases of peritoneal mesothelioma up to the end of 2000.7 Male ABA workers had very heavy levels of asbestos exposure; median cumulative exposure was 6 f/ml-years although the duration was short at a median of 4 months.8 Similarly, fewer deaths from malignant mesothelioma are predicted to occur among the Wittenoom women compared with the male asbestos workers. de Klerk et al3 predicted 692 cases between 1987 and 2020 and Berry4 estimated 255–654 cases for the same period. Follow-up by Berry et al in 2004 showed that the model that estimated 255 deaths to 2020 was the one that most closely described the observed mesotheliomas to the end of 2000. Using this model he predicted that 110 deaths from mesothelioma would occur between 2001 and 2020.7

These predictions may be overestimates for two reasons. The first concerns the age at first exposure to asbestos. When first exposed to asbestos at Wittenoom, 41% of the Wittenoom women were aged 14 years or younger.11 We have earlier reported a mesothelioma mortality rate for those aged 14 years or less at first exposure of 47 per 100 000, which was half that of those aged 15 years and older at first exposure who had a rate of 112 per 100 000.17 However, the lifetime risk for mesothelioma for those exposed at younger ages may well be higher. Age at first exposure was not incorporated into the mesothelioma rate models.

Secondly, we predicted mortality from mesothelioma in women who were not known to be dead at the end of 2004. This number included 27% who were lost to follow-up, most not having been traced since the date they left Wittenoom. Median cumulative exposure (2.02 f/ml-years) and median duration of residence (307 days) was lower among those lost to follow-up compared to those who were still alive (2.98 f/ml-years and 548 days) or dead (2.63 f/ml-years and 396 days) in 2004. More ABA workers (26%) compared to residents (21%) were lost to follow-up. It is not impossible that many of these women may already be dead, with some having died from mesothelioma. Ascertainment of mesothelioma is very good in Australia and in Western Australia particularly, but we have no knowledge about those women who may have changed their name in a state other than Western Australia or who may have subsequently departed from Australia.

There is also potential for some underestimation of future cases of mesothelioma. The projected downward trend in all cause mortality after 2004 would lead to a larger at-risk population and therefore larger expected numbers. Clements et al found that using current mortality gave predictions 16% less than using a projected life table.33 Their predictions extended to 2060 whereas we predict only to 2030, so the difference would be considerably less.

Forty women have died from malignant mesothelioma of the pleura from among approximately 3000 women and girls exposed to blue asbestos at Wittenoom between 1943 and 1992. We predict that there will be at least another 66 to at most 87 more mesothelioma cases among these women to 2030. The legacy of blue asbestos mining in Western Australia will continue to exact a high price well into the future.


We would like to thank Jan Sleith, Janice Hansen, Nola Olsen, Enzo Merler, Jim Leigh (Australian Mesothelioma Register), Tim Threlfall (Western Australian Cancer Registry), the JEM Foundation and the National Health and Medical Research Council.



Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.